Business paradigm for the AI world

Artificial Intelligence and Automation hold enormous promise for radical business transformation. Yet, as organizations embark on this journey, they encounter failures more often than not. Most companies start off with a pilot AI and Automation and experience initial gains. But as they commit more funds and try to scale up their efforts , they learn to their disappointment that the much promised productivity gains are modest and start tapering off much sooner than expected. On the other hand, there are a handful of companies that have managed to attain dramatic improvements. Companies like Uber and Airbnb have used AI successfully to upend traditional business models and disrupt the industry. What is the reason behind this dichotomy? What are the laggards doing wrong and how should they adapt to realize the true potential of AI and Automation ?

The primary reason why many companies are struggling to derive the promised benefits is their incremental approach i.e. they are just looking at existing business processes and trying to find pockets of opportunities for automation. Looking to automate current business processes can take you only thus far. We need to realize that the current business processes are a product of the Industrial Era. The stark truth is that these processes have already been optimized over the years and the room to gain further efficiencies is limited. When companies try to wring the last bit of efficiency from existing business processes, the benefits tend to taper off very soon. It is like trying to make lemonade out of already squeezed lemons. Only a few residual drops can be expected before bitterness quickly sets in. 

The new business paradigm for the AI world

In order to reap the benefits of AI, business processes need to be totally reimagined instead of resorting to incrementalism by automating existing business processes. That is precisely how AI driven companies like Uber and AirBnB are disrupting traditional businesses. Instead of viewing business processes as stepwise, standardised, repeatable processes, they need to be re-imagined as fluid and adaptive business processes. So, what is a fluid or adaptive business process all about? This can be best illustrated by the example of the evolution of digital maps. When online maps replaced paper maps, it was largely a digitized version of the paper maps. Although these were a big improvement over the paper maps, they lacked the adaptiveness of the current day maps which take into consideration GPS and traffic data. They not only suggest route options from point A to point B but also optimize the route for traffic. If the traffic situation changes, they dynamically show the new shortest route. The present day digital maps are thus adaptive and fluid as they dynamically change the route depending on traffic data. 

Similar to the data driven digital maps being replaced by static online maps, standardized business processes and rigid assembly lines are giving way to new types of business processes that continuously adapt on the fly to new data and market conditions. If Business processes have to benefit from AI, they need to adapt this new paradigm by moving away from the traditional standardized repeatable paradigm to a fluid and adaptive paradigm that is driven by dynamic data rather than static step wise traditional business process flows. The data could be personalized, real time, contextual. Such data driven business processes are personalized and contextualized for the needs of the individual rather than being an efficient once size fits all process better suitable for mass production. 

The traditional maintenance processes are great examples of archaic processes that are crying for change. The processes are also often a well laid out trap by manufacturing companies to milk their customers through lucrative annual maintenance contracts. After I bought my Maruti car, I felt like a golden goose that laid an egg for Maruti every 6 months when I went to service my car. It is not as if machines don’t need maintenance. It is just that a lot of money is unnecessarily spent on checks that are not warranted while the checks that are really needed are often missed out. Just a week after my regular car maintenance, my battery gave away and I was stranded in the middle of the highway. In short, maintenance needs to be far more contextual and precise rather than a mindless routine at fixed intervals (designed to be a steady source of cash flow)

GE reinvents the age old maintenance process

GE’s has reinvented the age-old maintenance processes using AI driven predictive maintenance services using contextual data. GE uses the concept of “digital twin” for predictive maintenance of sophisticated machinery like jet engines and gas turbines. A digital twin is built by using data to represent and model the machine. Data is collected from the physical machine using sensors and the data collected include parameters like heat, vibration and noise. The data is pushed to the cloud and sophisticated machine learning algorithms analyze patterns in the data and predict the need for maintenance, ultimately reducing  the probability of unplanned downtime in that machine. The advantages are twofold. Firstly, it prevents costly downtimes. Secondly, it also reduces costs by avoiding routine maintenance and maintenance would be done on ‘need basis’ and that too on  specifically identified machine parts.

Stich Fix personalizes apparel business

Another great example of a company that has reinvented age old business processes for the new AI world is Stitch Fix. Stitch Fix is an online apparel company that has completely reinvented the apparel business model through personalization at scale. Traditional online fashion companies rely on personalized recommendations through cross selling and upselling recommendations. The conversion in these kind of personalization is low. Stitch Fix instead uses data right upfront by having users provide their details and personal choices while signing up. Customers do not have to order anything. Based on the details and personal choices provided by the customers, Stitch Fix ships a box of five pieces of clothing along with  accessories tailored to the personal choices of the customers. Customers can keep what they like and return the rest. Within just eight years after it was founded, the company made annual revenues of over USD 1.5 Billion dollars and is now valued at over 2 Billion dollars. Conventional business has been turned upside down. Shipments are made even without ordering. Data isn’t just woven into the company’s culture; data actually defines its culture. Personalization is not based on cross selling and upselling but  is in-built in their process of capturing data filled by prospective customers thereby enabling the company to ship apparel tailored to their individual choices. 

In summary

For businesses to succeed in the AI world, the archaic ways that are a byproduct of the Industrial era must be relinquished. The old paradigm of traditional business processes have been standardization, repeatability and efficiency and these are suitable only for mass production. Furthermore, efforts to automate the existing business processes any further will only lead to minor and incremental productivity gains that will soon taper off.

It is imperative that the traditional processes are re-imagined for the AI world. These re-imagined business processes need to be fluid and adaptive. They should shun ‘one size fits all’ approach and tailor their business outcomes in a personalized and contextualized manner, often responding to the market conditions dynamically in real time.